OPINION: Rapid progress is being made in the field of artificial intelligence (AI) and robotics. Milestones are being passed in areas as diverse as transcription (computers now outperform humans at transcribing spoken Mandarin), diagnosis (computers outperform the best doctors at diagnosing pulmonary disease) and warfare (computers outperform the best human pilots in air to air combat).
Most experts in AI estimate it will take at least 50 years to get to human-level intelligence in machines. A serious research effort in "AI safety" has begun to prepare for this moment and ensure the goals of any such intelligent or super-intelligent machines align with those of humanity. Fears that the machines will take over remain more the concern of Hollywood than the laboratory.
Australia is close to the front of this revolution and punches above its weight in AI research. The impact AI will have on society will likely be felt early on in Australia compared with many other developed countries. We will not have the luxury of observing what happens in the US or elsewhere. We will need to lead the way in adapting to the changes and challenges.
Among these challenges are issues such as protection of privacy and transparency as more decisions are handed to machines. Closely connected to concerns about transparency are issues around trust. How do we know when to trust a machine? When we observe a computer performing intelligently on one problem, we often tend to suppose it will work equally well on another. However, AI remains brittle. Our smart computers can be surprisingly dumb when the problem changes even slightly.
In safety and security-critical areas, there are already well-developed tools and techniques for verification and validation of computer systems. Yet despite what high-tech companies such as Google might have us believe, algorithms – especially those using Machine Learning – can be biased. Algorithmic discrimination will start to trouble society increasingly. If we are not careful, many of our hard-fought rights against racial, religious, sexual, age and other types of discrimination will be lost to machines that are not transparent and that we should not trust.
Other aspects of our society will also be affected by AI. We are already witnessing the impact of algorithms on politics and political debate through data-driven political marketing while globally there is an arms race under way today to develop "killer robots".
This is not the first technological revolution that has affected society so we might look for lessons that can be learnt from history. Perhaps the closest parallel is the Industrial Revolution. At the time of the Industrial Revolution, the world took several large shocks that helped society adapt to the change. Two world wars and the intervening Great Depression led to an unusual reversal in inequality. We might expect equally large societal changes will occur for the coming AI revolution. A worrying lesson from history is there was about 50 years of pain at the start of the Industrial Revolution during which prosperity for many in society went backwards.
These ethical concerns and historical lessons have implications for government. All concern education in one way or another. This is because education is one of the most important and powerful tools at our disposal in adapting to the coming changes.
In Australia and the US, a major problem within the field of computer science, and especially within artificial intelligence, is the 'sea of dudes' problem.
In handing over many of our decisions to machines, we will need to make explicit in computer code many of our society's ethical choices. A citizenship educated in ethics, society and civics is therefore essential. The education system needs to prepare us for this future of "computational ethics". In time, machines will become as creative and adaptable as humans, but they cannot speak to the human experience: about love, death and all the things that make us unique. It follows that creativity can and should be taught more actively.
At present our education system focuses on lifting cognitive abilities. However, in some countries attention is also given to improving emotional intelligence. This would be a good idea in Australia.
For many, education stops when they leave school or university. Ultimately, just as the Industrial Revolution made it essential that universal education was provided to the young, the AI revolution will make it essential that education is provided to people at every age of their lives.
In Australia and the US, a major problem within the field of computer science, and especially within artificial intelligence, is the "sea of dudes" problem. The under-representation of women in AI and robotics is undesirable for many reasons. Women will be disadvantaged in an increasingly technically focused job market. But it may also result in AI systems that fail to address issues relevant to half the population, and even to systems that perpetuate sexism. More initiatives are needed to get young girls interested in science, technology, engineering and maths (STEM) in general, and AI and robotics in particular.
Providing one robot per child could be one solution. There is evidence that access to robots, especially at an early age, can help bring girls into STEM. Additionally we need citizens to understand the fundamental principles of computation. If we don't, a large section of the population will be greatly disadvantaged as much technology will simply be magic to them. Robots will offer an excellent platform on which to teach such computational thinking.
Critically we also need a government-wide report on how to prepare for the changes that AI and robotics will bring to society. These are technologies that will touch almost every aspect of our lives. We need to start preparing for this future. There are many ethical challenges ahead, ensuring that machines are fair, transparent, trustworthy, protective of our privacy and respect many other fundamental rights. Education is likely to be one of the main tools available to prepare for this future.
Toby Walsh is the Scientia Professor of Artificial Intelligence at UNSW Sydney. This is an edited extract from Future Frontiers: Education for an AI World, a NSW Department of Education publication to be launched on Thursday at the Education for a Changing World Symposium in Sydney. This article was originally published in the Sydney Morning Herald.